Skip to main content

Harmony Search is evolving: what we’ve delivered and what’s coming next

  • September 25, 2025
  • 13 replies
  • 508 views
A
Docebian

Thank you all for the feedback, suggestions, and the interest you’ve shown in this feature and initiative. The information you’ve shared, and continue to share, is essential and extremely valuable to us.

I’d like to take this opportunity to give you an update on what we’ve delivered since the release, what we’re currently working on, and what we’re planning next.
 

Recent updates

 

Harmony in “All Results”

Harmony can now provide answers not only in the conversational interface but also directly on the All Results page, when the query type allows for a potential Harmony response. In these cases, the answer will appear as the very first item at the top of the results list.


 

Vertical Search

We’ve released a new feature called Vertical Search. This allows users to pre-filter the content Harmony uses to generate its response. Instead of retrieving information from all content visible to the user, Harmony will only reference the selected sources. Supported content can be selected either from the conversational interface, from the results page, or from the suggested results dropdown via the Ask icon. This option also excludes the model’s generic knowledge from responses.
 

 

Upgraded LLM

We’ve updated the LLM that powers Harmony. It’s now faster, produces higher-quality answers, and its generic knowledge has been refreshed with a more recent cutoff date. We’re also working on making response rendering faster.
 

SCORM Support Enhancements

We’ve deployed a new version of our distiller in production. This enables transcript generation and ensures newly uploaded SCORM packages created with Articulate Rise or Articulate Storyline are usable by Harmony. By mid-October, we plan to reprocess all previously uploaded SCORM packages, bringing accurate transcripts to around 70% of all existing SCORMs. This will not only improve their ranking and relevance in traditional search but also enrich Harmony’s dataset for conversational responses.
We don’t intend to stop here, we’re already exploring ways to expand transcription coverage to SCORMs created with other authoring tools (we currently have good coverage for Go1 and OpenSesame) as well as other formats, such as Tin Can/xAPI.

 

What’s next
 

We’ve heard your feedback loud and clear: administrators will soon have the option to disable the model’s generic knowledge in Harmony’s conversational answers.

Following that, we plan to introduce another option allowing Harmony to include free catalog content (visible to the user but not yet actively enrolled) in its responses, an improvement that will benefit several use cases.
 

Looking further ahead
 

Beyond these immediate priorities, we’re also evaluating and preparing additional initiatives, including:

  • A search analytics dashboard to give administrators a clear overview of user search data within the platform and other information.
  • Federated search by integrating our search engine with third-party systems such as Google Drive and SharePoint.
  • Enhancements to the standard search system with updated technology.
  • Support for searching content marketplace items that haven’t yet been imported into the platform.

And more...

We’re actively collecting feedback to guide further enhancements.
Thank you again for all your valuable input.

    13 replies

    martiniwinski
    Influencer II
    Forum|alt.badge.img
    • martiniwinski
    • Influencer II
    • September 25, 2025

    These updates are great so far!

    Another nice feature would be to for Harmony to provide links to the resources (SCORM, etc.) it referenced for its output - whether inline within the paragraph or below the response.

    Martin Iwinski
      D
      Influencer I
      • DanielM
      • Influencer I
      • September 25, 2025

      All good news! And my favorite is the option to disable generic knowledge. Separating the generic knowledge - regardless how it is implemented - is essential to show users answers from the training they buy. Thanks!

      Daniel
      A
      Docebian

      These updates are great so far!

      Another nice feature would be to for Harmony to provide links to the resources (SCORM, etc.) it referenced for its output - whether inline within the paragraph or below the response.

      Hi ​@martiniwinski, thank you for your feedback. This information should already be available right after Harmony Search provides a response, as long as the answer is generated using platform resources rather than the model’s general knowledge. SCORM packages will also be referenced once supported and used for answer generation.

       

        Jessica Tart
        Helper II
        Forum|alt.badge.img+4
        • Jessica Tart
        • Helper II
        • September 25, 2025

        These updates are great so far!

        Another nice feature would be to for Harmony to provide links to the resources (SCORM, etc.) it referenced for its output - whether inline within the paragraph or below the response.

        Hi ​@martiniwinski, thank you for your feedback. This information should already be available right after Harmony Search provides a response, as long as the answer is generated using platform resources rather than the model’s general knowledge. SCORM packages will also be referenced once supported and used for answer generation.

         

        Will Harmony be able to suggest a course and provide a link to the course for the user to take? For example, the user searches for Travel Policy and Harmony provides a link to the Travel Policy eLearning that the user isn’t enrolled in. 

        Jessica Tart
        ebosworth
        Novice II
        • ebosworth
        • Novice II
        • September 26, 2025

        Love all the updates - very helpful and we are looking forward to what is in the works for the future.

        However, we don’t use Storyline or Rise, so I’m curious to know more about the timeline for parsing SCORM files from other authoring tools. We use Evolve from Intellum for our SCORM files. 

        If authoring tools have their own specific packaging logic, I’m guessing that Docebo did an publisher specific parser for Articulate products. 

        1. Is the plan is to do more publisher specific parsers or are you moving towards a model that can parse e-learning content from any publisher?
        2. If Docebo is going the publisher specific route, what criteria is Docebo going to use for selecting additional publishers?
        A
        Docebian

        These updates are great so far!

        Another nice feature would be to for Harmony to provide links to the resources (SCORM, etc.) it referenced for its output - whether inline within the paragraph or below the response.

        Hi ​@martiniwinski, thank you for your feedback. This information should already be available right after Harmony Search provides a response, as long as the answer is generated using platform resources rather than the model’s general knowledge. SCORM packages will also be referenced once supported and used for answer generation.

         

        Will Harmony be able to suggest a course and provide a link to the course for the user to take? For example, the user searches for Travel Policy and Harmony provides a link to the Travel Policy eLearning that the user isn’t enrolled in. 

        Hi ​@Jessica Tart, yes. We are exploring the possibility of adding an option that allows Harmony to include free catalog content (visible to the user but not yet actively enrolled) in its responses. The idea is that when a user asks Harmony about something like the Travel Policy, if the content type is supported, Harmony will provide an answer along with a link to the relevant course catalog page, allowing the user to self-enroll. As soon as we have more details, we will share further information on this topic.

          A
          Docebian

          Love all the updates - very helpful and we are looking forward to what is in the works for the future.

          However, we don’t use Storyline or Rise, so I’m curious to know more about the timeline for parsing SCORM files from other authoring tools. We use Evolve from Intellum for our SCORM files. 

          If authoring tools have their own specific packaging logic, I’m guessing that Docebo did an publisher specific parser for Articulate products. 

          1. Is the plan is to do more publisher specific parsers or are you moving towards a model that can parse e-learning content from any publisher?
          2. If Docebo is going the publisher specific route, what criteria is Docebo going to use for selecting additional publishers?

          Hi ​@ebosworth, and thanks a lot for sharing this information. It’s really valuable in helping us understand how and where to plan our next developments.

          Right now, we’re following a straightforward strategy based on usage volumes: the more an authoring tool is used by our customers, the higher the priority we give to supporting transcript indexing in Harmony Search, so we can meet as many requests as possible.

          That said, this applies only to prioritization. Our broader goal is still to extend support to as many authoring tools as possible, so we really appreciate your feedback.

          We can’t confirm short-term support for SCORM files from Evolve just yet, but I can assure you that we’ll review the request and share an update as soon as we have more information.

          ebosworth
          Novice II
          • ebosworth
          • Novice II
          • September 29, 2025

          Love all the updates - very helpful and we are looking forward to what is in the works for the future.

          However, we don’t use Storyline or Rise, so I’m curious to know more about the timeline for parsing SCORM files from other authoring tools. We use Evolve from Intellum for our SCORM files. 

          If authoring tools have their own specific packaging logic, I’m guessing that Docebo did an publisher specific parser for Articulate products. 

          1. Is the plan is to do more publisher specific parsers or are you moving towards a model that can parse e-learning content from any publisher?
          2. If Docebo is going the publisher specific route, what criteria is Docebo going to use for selecting additional publishers?

          Hi ​@ebosworth, and thanks a lot for sharing this information. It’s really valuable in helping us understand how and where to plan our next developments.

          Right now, we’re following a straightforward strategy based on usage volumes: the more an authoring tool is used by our customers, the higher the priority we give to supporting transcript indexing in Harmony Search, so we can meet as many requests as possible.

          That said, this applies only to prioritization. Our broader goal is still to extend support to as many authoring tools as possible, so we really appreciate your feedback.

          We can’t confirm short-term support for SCORM files from Evolve just yet, but I can assure you that we’ll review the request and share an update as soon as we have more information.


          Thank you Alessandro!

          We’re also VERY interested in the Search Analytics Dashboard you mentioned was coming in the future.

          • A search analytics dashboard to give administrators a clear overview of user search data within the platform and other information.

          We can’t get Google Analytics to deliver on providing specifics on what our users are searching for right now which is really important to us. Any high level idea of when that might be ready?  Is that an end of 2025 goal or more like a 2026 goal? 

          - Elke

          A
          Docebian

          Love all the updates - very helpful and we are looking forward to what is in the works for the future.

          However, we don’t use Storyline or Rise, so I’m curious to know more about the timeline for parsing SCORM files from other authoring tools. We use Evolve from Intellum for our SCORM files. 

          If authoring tools have their own specific packaging logic, I’m guessing that Docebo did an publisher specific parser for Articulate products. 

          1. Is the plan is to do more publisher specific parsers or are you moving towards a model that can parse e-learning content from any publisher?
          2. If Docebo is going the publisher specific route, what criteria is Docebo going to use for selecting additional publishers?

          Hi ​@ebosworth, and thanks a lot for sharing this information. It’s really valuable in helping us understand how and where to plan our next developments.

          Right now, we’re following a straightforward strategy based on usage volumes: the more an authoring tool is used by our customers, the higher the priority we give to supporting transcript indexing in Harmony Search, so we can meet as many requests as possible.

          That said, this applies only to prioritization. Our broader goal is still to extend support to as many authoring tools as possible, so we really appreciate your feedback.

          We can’t confirm short-term support for SCORM files from Evolve just yet, but I can assure you that we’ll review the request and share an update as soon as we have more information.


          Thank you Alessandro!

          We’re also VERY interested in the Search Analytics Dashboard you mentioned was coming in the future.

          • A search analytics dashboard to give administrators a clear overview of user search data within the platform and other information.

          We can’t get Google Analytics to deliver on providing specifics on what our users are searching for right now which is really important to us. Any high level idea of when that might be ready?  Is that an end of 2025 goal or more like a 2026 goal? 

          - Elke

          ​@ebosworth More realistically, the Search Analytics Dashboard will be an initiative for 2026.

            Dominik
            Contributor II
            • Dominik
            • Contributor II
            • October 1, 2025

            These are great news! I’ve really enjoyed using Harmony Search so far, and with the upcoming enhancements, it’s getting even better.

            We primarily use Articulate products to create our training content and export it using xAPI.

            So far, Harmony’s performance with xAPI content has been mixed. For example, when using a Rise course, it seems that the transcript is only partially extracted. Harmony can provide answers for certain chapters, but not for others within the same course.

            Will the xAPI support in this area be improved as well, as suggested in the original post?

            Thank you.

            A
            Docebian

            These are great news! I’ve really enjoyed using Harmony Search so far, and with the upcoming enhancements, it’s getting even better.

            We primarily use Articulate products to create our training content and export it using xAPI.

            So far, Harmony’s performance with xAPI content has been mixed. For example, when using a Rise course, it seems that the transcript is only partially extracted. Harmony can provide answers for certain chapters, but not for others within the same course.

            Will the xAPI support in this area be improved as well, as suggested in the original post?

            Thank you.

            Hi ​@Dominik , thanks for your feedback! We’re aware that xAPI support could be improved, and I can confirm that we plan to work on this by the end of the year. We don’t have a final completion date yet, but this is something we’ve seen come up often, and we’re committed to making the xAPI packages generated with Articulate more reliable and improving their transcript extraction.
            We really appreciate your input, it helps us prioritize what matters most to users!

              Dominik
              Contributor II
              • Dominik
              • Contributor II
              • October 1, 2025

              Thank you for your prompt feedback ​@Alessandro Ortolina. Sounds great.

                N
                Novice II
                • ncdocebo
                • Novice II
                • December 19, 2025

                Would like to see Harmony include more result types, for instance rather than navigating to the custom reports builder page, I would like to use search to type in the name of the report and the option appear. Other result types like users, power users, channels, catalogs, notifications, summary/custom reports, events, etc.

                  Terms & ConditionsCookie settingsAccessibility statement